Re: names using git config

Ryan [off-list ref] writes:

On Thu, Jun 25, 2009 at 10:41 AM, Junio C Hamano[off-list ref] wrote:

quoted

Ryan [off-list ref] writes:

quoted

Hi,
  Thank you for your response.

1) git config user.name 'Matthieu M. Moy' Works
2) git config user.name 'Matthieu M.'  does not

Name cannot end in a '.' <dot>. in git config user.name

That is correct, but I think we are a bit too strict in sanitizing names.
The code is ident.c::copy() whose intent was to sanitize useless runs of
commas we slurp from GECOS field, but the sanitization is done also for
anything specified by the end user.

What are you suggesting me to do? Apply the below patch.
I may want to submit patches to the linux kernel. Is this okay?

Sorry if I was unclear.  I wasn't suggesting anything in particular to
you.  The only concrete thing addressed to you in my message was the
answer "That is correct" and all the remainder of my message, including
the patch, was primarily meant as illustration for other git developers to
think about.

I do not think this matters that much in practice, as long as you are
submitting patches via e-mail, the usual method for leaf-level developers.
The maintainers apply your patches under the name your e-mail's From:
header carries, not your user.name.

If you apply the patch to your copy of git, your "git log" output will
start showing otherwise forbidden names that have leading or trailing dots
(and other letters like semicolons) because your modified copy of "git
commit" will allow them, and I think "git commit -s" also will pass the
trailing dots intact, but that is about the extent of the damage.  Without
applying the patch, you will perhaps see your "Signed-off-by: Foo M
[off-list ref]" line without the trailing dots after the name, but
that is something you can easily modify in your MUA when you are sending
the e-mail out.